|      
     
    home
 
    overview
 
    features
 
    download
 
    manuals
 
    publications
 
    FAQ
 
    contact
 
 
 
       last update:15|03|2007
 
 

The most distinguishing characteristics of Inputlog to date are:

  • word processor independent functionality
  • the standard XML structure of the output
  • the logging of speech recognition
  • parsing technology
  •  

     word processor independency

     

    Inputlog registers every keystroke and mouse movement independently of the word processor used. Inputlog is designed to log and analyze writing data produced in Microsoft Word. However, the program also logs keyboard and mouse actions in other Windows based programs. In other words, not only writing processes as such can be observed with Inputlog, but also basic processes like consulting websites or programming in any Windows based language.

     XML structure of output files

     

    Inputlog 2.0 integrates the universal XML structure. This allows users to define the tags (markup) that are needed to identify the data and text in XML documents. The advantage of XML is that researchers can easily adapt the data to their own research needs. Research data can be built in the same way as the Inputlog output and can then be easily integrated into one another. The picture shows how the XML structure of Inputlog is built.

    Each states the information between the tags: e.g. writing mode 2 is a mouse movement or click. XML structured data of various programs can be easily merged because of the markup language.

     speech recognition

     

    Inputlog is currently the only logging tool that can integrate the input of dictation devices using speech recognition software. For this reason we chose the most widely used speech recognition software: Dragon Naturally Speaking. The specially designed logging add-on in the speech recognizer (in combination with a Python script) enables us to integrate the dictated text with the data logged by Inputlog. Comparable to the general logging file generated by Inputlog, the logging file of Naturally Speaking relies on timestamps. Via a Python script that was developed for this purpose, we generate an XML file of a recorded speech session that is basically structured in the same way as an Inputlog data file. The result is a single file that can be used for further analysis of multimodal writing sessions in which speech input is combined with keyboard and mouse.

    This is a short XML excerpt of a logged writing process in Microsoft Word in which three writing modes were used (column 1: writing mode): keyboard (1), mouse (2), and speech (3 - dict).

     parsing

     

    Generally speaking, parsing refers to the process of analyzing an input sequence. The processing of input data depends on what has preceded, and even on what follows. As such, keeping track of these dependencies in order to verify how to process the data is called parsing. Inputlog uses the tools Flex and Bison to implement the The parsing techniques simplify the overall program by separating the input and processing components and by providing a natural, modular structure. The main advantages of parsing are:

  • A more readable program structure.
  • Faster analyses because of efficiency structure.
  • Easy maintenance, modification and elaboration because of structure and modularity.